Identification of Ozone Sources in the Western US Dan Jaffe, University of Washington. Mt. Bachelor Oregon 2.8 km or 9000 above sea level

Transcription

1 Identification of Ozone Sources in the Western US Dan Jaffe, University of Washington Mt. Bachelor Oregon 2.8 km or 9000 above sea level

2 Acknowledgements

3 The Challenge: New O 3 Standard in Light of Increasing Background Most urban areas in the West have made substantial progress in reducing peak O 3 mixing ratios. But the new O 3 standard presents our greatest challenge yet. Springtime background O 3 is increasing; Wildfires are increasing. Need to improve our understanding of O 3 sources. Need to improve our methods to document which O 3 is controllable, and which is not, on a day-to-day basis.

4 Los Angeles, Max Daily 8-hour Average O 3 New standard AQS site #

5 O 3 Source Local photochemical buildup Sources of O 3 in the Western US Meteorological Characteristics Stagnation, high temps. Chemical characteristics CO/NOx/VOCs/PM consistent with local sources Controll -able Y Regional transport (domestic sources) Regional transport from major source regions (eg Calif) *Currently not well characterized CO/NOy/VOCs consistent with upwind sources+chemistry Y Upper trop/lower strat intrusions (UTLS) Post-cold front Broad spatial distribution (high O 3 in non-urban areas) Very dry air. N Very long-range transport (VLRT) Important at higher elevation. Subsidence and mixing into the boundary layer can enhance local concentrations. Dry. Can be hard to distinguish from UTLS without good chemical data. N Wildfire smoke Warm. Can be stagnant or not. Can be regional or large distant fires. Chemistry complex and diff from typical urban. O 3 enhancements not always seen. O 3 -PM often poorly correlated. PM/CO/NOy always well correlated and ratios very diff from typ urban. N

6 How to id sources of O 3? Observations: Spatial relationships; Tracer ratios; Meteorology/trajectories; Satellite observations. Modeling: Regional Eulerien modeling; Source apportionment modeling (e.g. WRAP-DEASCO3; Statistical modeling.

7 Mt. Bachelor, Oregon, (MBO) 2.8 km asl The only high elevation/free trop research site on west coast of U.S. Continuous observations of CO, O 3, aerosols and Hg since 2004; Frequent detection of Asian pollution and biomass burning plumes; More than 2 dozen papers since 2004 on O 3, PM. Hg, LRT, wildfires, etc. Key goal: Identify importance of global sources on US air quality.

8 Diurnal circulation pattern at Mt. Bachelor h ν Day: upslope flow brings modified BL air to summit. This air is wet and usually low in O 3. Night: downslope flows brings Free Tropospheric (FT) air to the summit. This air is dry and usually high in O 3. ID of Free Tropospheric Air Time of day. Water vapor mixing ratio Chairlift soundings, observations of NOx (Weiss 2006, 2007; Fischer 2009; 2010; Reidmiller 2011)

9 MDA8 O 3 at MBO New U.S. standard of 70 ppbv! Current U.S. O 3 standard Max daily 8 hour avg O 3 at Mt. Bachelor for Latest 3-year design value =79 ppbv What are causes of high O 3 at MBO and how does this influence surface AQ?

10 Causes of high O 3 days at MBO: Strat intrustions, Asian pollution and wildfires Ambrose et al 2011 Baylon et al 2014

11 Role of background O 3 on urban air quality Wigder et al 2013

12 Modeled FT O 3 tracer for 4/6/2010 (high O 3 day) using CMAQ w/12 km grid Vertical crosssection Surface mixing ratio MBO Boise MBO Boise Wigder et al 2013

13 Springtime O 3 trend at MBO: Gratz et al, 2014

14 Changes in Spring O 3 at MBO 95 th percentile 50 th percentile 5 th percentile This is likely due to the build up of Asian emissions (Gratz et al 2014).

15 Changes in Spring O 3 at MBO 95 th percentile Spring th percentile 5 th percentile This is likely due to the build up of Asian emissions (Gratz et al 2014).

16 MBO MDA8s May O 3 5/29 Asian LRT (High CO+O 3 ) 5/ /9 UTLS (High O 3 low CO and RH) Was high O 3 seen in 2012 at surface monitoring sites? 16

17 Importance of spatial correlation in Western US Selected O 3 monitoring sites SJV

18 Location May 2012 Relative to May (MDA8, ppbv) Great Basin NV (2.1 km) Mt Bachelor OR (2.8 km) Boise ID MSA (0.8 km) Reno-Sparks, NV (1.5 km) Salt Lake City, UT (1.4 km) +7.0 Tehama Cty CA (0.6 km) Portland OR MSA (0.2 km) Lassen N.P. CA (1.8 km) Siskiyou Cty CA (0.8 km) San Joaquin Cty (0.1 km) Higher elevation sites shows greater enhancement.

19 Spatial correlation between sites in W. US, Spring 2012 Plotted with 3 day smoothing of MDA8 values This spatial-temporal correlation tells us that on these dates, high O 3 was not locally generated.

20 Spatial correlation between Ozone sites in W. US, MDA8 O 3 for Spring 2012 with 3-day Smoothing Baylon et al 2015

21 Wildfires Wolverine Fire, Lake Chelan, Washington Aug 3, 2015

22 Area Burned for US Wildfires (NIFC) The last decade has seen a significant increase in the area burned. Approx 70% of these fires are in the Western US

23 Summary of ΔO 3 / ΔCO from >100 published studies Boreal/ Temperate: Plume Age Mean O 3 / CO (ppbv/ppbv) (# plumes) Range of O 3 / CO 1-2 days (n=55) days 0.15 (n=39) days 0.22 (n=29) Tropics/ Subtropics: Plume Age Mean O 3 / CO (ppbv/ppbv) (# plumes) 1-2 days 0.14 (n=59) days 0.35 (n=13) days 0.63 (n=18) Range of O 3 / CO Jaffe, D.A. and Wigder, N.L., Ozone production from wildfires: A critical review. Atmos. Envir., doi: /j.atmosenv , Jaffe & Wigder (2012)

24 Wildfires can make O 3 very quickly O 3 CO Aerosol scattering Mt. Bachelor observations of the Pole Creek Fire on three successive days. O 3 production of ppbv in 6 hours. Many other examples of fast O 3 production in lit.

25 Median O 3 summer MDA8 in Salt Lake City, Utah Large fire years Low fire years Wildfires are responsible for year to year variation in frequency of high O 3 days.

26 Median MDA8 in SLC, CASTNET, OC, AIRS CO and AOD R values between 0.58 and 0.86 for these. Wildfires are partly responsible for year to year variation in frequency of high O 3 days (Jaffe et al 2013).

27 More tools

28 Observations Observations of key tracers and meteorology are critical to understand causes for high O 3 days; Tracer ratios provide more information than one pollutant alone; Most agencies follow EPA specs for O 3 and PM 2.5 and these are done reasonably well; Not all regulatory sites have meteorology (!); At minimum, need good quality measurements of CO and NOx. VOCs and speciated PM are also highly desirable. A specific tracer of fires (such as CH 3 CN) is also very useful. Shameless self promotion my group is working on a simple, transportable suitcase method to measure CH 3 CN.

29 PM2.5-CO relationship in smoke at some NCORE sites, hourly data SLC, Utah NCORE site, Aug 7-14, Smoke confirmed using multiple models (Jaffe 2013). Spokane, WA site Aug 26, 2015 Smoke confirmed. Spokane, WA site Aug 21, Smoke confirmed.

30 Examples of PM2.5-CO relationship at Mt. Bachelor Observatory-hourly obs Smoke plume observations at the Mt. Bachelor Observatory, Sept 9, 2014 using Picarro G2302 Smoke plume observations at the Mt. Bachelor Observatory, Sept 25, 2009 using Thermo 48C TLE. Our observations show that fire plumes have ΔPM/ΔCO ratios of µg/m 3 per ppb.

31 Fires vs typical urban emissions Typical urban obs EPA emission inventory Fire emission inventory (Akagi 2011) Fire plume observations NOy/CO molar PM2.5/CO μg/m 3 per ppb to to Observations of CO, PM2.5 and NOy can distinguish source of pollutants on high O 3 days.

32 Eularian modeling vs Statistical modeling Eularian modeling: Gridded emissions, meteorology, solar fluxes (J values). Use known photochemistry and transport to model mixing ratios. For wildfires significant challenges with emissions, plume rise, aerosols and the chemistry, which can be very different from typical urban photochemistry. Modeled concentrations may differ significantly from observations making quantitative attribution difficult. Statistical modeling: Examines the relationship between the observed mixing ratios and other factors. Possible factors to include are temp, wind speed, RH, solar flux, etc. Outliers (high residuals) represent an additional O 3 source and are candidates for further investigation. O 3 = A*temp + B*winds + C*DOY + residual (Jaffe et al 2004; 2013; Camalier et al 2007; CARB 2011; EPA 2015)

33 SLC MDA8 vs Daily max T (SLC airport) R 2 = 0.45

34 R 2 = 0.28 SLC MDA8 vs DOY

35 Statistical model for SLC: R 2= 0.60 SLC Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta Constant T-max Daily Avg Wnd Spd Yr DOY Parameter inclusion requires: 1. Statistical significance 2. Reasonable physical interpretation

36 Why use Statistical Models? 1. Give important information on meteorology impacts on AQ in your area; 2. Give information on usual relationship between met variables and AQ; 3. Provides quantitative information on unusual conditions; e.g. exceptional events. In Jaffe et al 2013 (EST) we compared statistical model with two Eulerien models for wildfire O 3 impacts in SLC, Reno and Boise.

37 EPA 2015: Guidance on the Preparation of Exceptional Events Demonstrations for Wildfire Events that May Influence Ozone Concentrations Miller D., DeWinter J., and Reid S. Documentation of data portal and case study to support analysis of fire impacts on ground-level ozone concentrations. Technical memorandum prepared for the U.S. Environmental Protection Agency, Research Triangle Park, NC by Sonoma Technology, Inc., Petaluma, CA, STI , Sept. 2014: Documentation of Data Portal and Case Study to Support Analysis of Fire Impacts on Ground-Level Ozone Concentrations

38 AirNow MetData

39 Comparison of my original statistical model with AirNow statistical model My model (Jaffe et al 2014-EST) AirNow model Location Salt Lake City, UT Salt Lake City, UT Time period % variance explained Key met variables How long to construct 0.60 (60% 0.59 (59%) Maxtemp, WS, DOY, 700 mb zonal wind, etc. Maxtemp, WS, 850 mb temp, etc. ~ 1 month ~ 1 hour

40 Comparison of residuals in my original statistical model with AirNow statistical model

41 AirNow MetData

42 Anna Wood (EPA-AQPD) and Chet Wayland (EPA-AQAD) Westar Fall 2015 meeting Final area designations due Fall based on final DVs (120-day letters by June 2017) Early-certified 2017 data may also be relevant to final designations. We expect state designation recommendations to be based on and preliminary 2016 data, including any exceptional event considerations. Exceptional event demonstration submission deadlines: October 1, 2016 for events May 31, 2017 for 2016 events

43 Anna Wood (EPA-AQPD) and Chet Wayland (EPA-AQAD) Westar Fall 2015 meeting Preliminary stakeholder discussion questions: From the stakeholder perspective, what additional data elements and/or model improvements are needed to better characterize background O3 levels across the U.S.? From the stakeholder perspective, has EPA properly characterized the various CAA provisions under consideration for areas influenced by background O3? Are there sufficient technical tools and data available to make the demonstrations necessary to invoke relevant CAA provisions?

44 More on process and timeline-tentative EPA plans to convene state agency/stakeholders meeting in Phoenix in late Feb (last week?) to discuss implementation; AWMA will convene a second webinar on O 3 implementation in spring (April or May?) Exceptional event designations for due Oct 1, Need a continuous dialogue. State agencies

45 Summary 1. The new O 3 standard will be a major challenge for the western US. While O 3 is improving in many urban areas the standards are now approaching background concentrations; 2. Stratospheric intrusions (UTLS) and very long-range transport can be identified from tracer ratios, met and spatial correlations; 3. Wildfires are becoming increasingly important. While the PM and health effects are obvious, there are many uncertainties around secondary aerosol and O 3 production; 4. Need to develop and use all the tools available to us. This includes better observations, tracer ratios, spatial-temporal correlations and statistical modeling; 5. Need to push EPA to improve these tools.